|Home | About | Journals | Submit | Contact Us | Français|
Direct reprogramming of fibroblasts into cardiomyocytes is a novel strategy for cardiac regeneration. However, the key determinants involved in this process are unknown.
To assess the efficiency of direct fibroblast reprogramming via viral overexpression of GATA4, Mef2c, and Tbx5 (GMT).
We induced GMT overexpression in murine tail tip fibroblasts (TTFs) and cardiac fibroblasts (CFs) from multiple lines of transgenic mice carrying different cardiomyocyte lineage reporters. We found that the induction of GMT overexpression in TTFs and CFs is inefficient at inducing molecular and electrophysiological phenotypes of mature cardiomyocytes. In addition, transplantation of GMT infected CFs into injured mouse hearts resulted in decreased cell survival with minimal induction of cardiomyocyte genes.
Significant challenges remain in our ability to convert fibroblasts into cardiomyocyte-like cells and a greater understanding of cardiovascular epigenetics is needed to increase the translational potential of this strategy.
Unlike the hearts of teleosts or zebrafish,1 the mammalian heart undergoes a fibrotic response with minimal regeneration.2 The inability of the adult heart to completely repair itself spurs the development of strategies to transplant endogenously or exogenously-derived cardiac cells into patients after ischemic injury3–8. These transplantation strategies have led to measurable successes in functional recovery or remuscularization. However, significant challenges remain regarding the efficiency, feasibility and efficacy of this approach6–9. Thus, strategies aimed at directly converting cardiac scar fibroblasts into cardiomyocytes are appealing as they circumvent problems associated with cell purity and survival after transplantation.
To this end, Ieda and colleagues reported that overexpression of Gata4, Mef2c and Tbx5 (GMT) could reprogram murine cardiac fibroblasts (CFs) and tail tip fibroblasts (TTFs) into cardiomyocytes in vitro.10 Furthermore, infected fibroblasts could survive and reprogram after transplantation into a murine heart. GMT-induced fibroblasts demonstrated gene expression profiles similar to mature cardiomyocytes and beat spontaneously in vitro. Conversion to cardiomyocyte-like epigenetic states was reported through the de-repression of histone markings at promoters of sarcomeric genes. These results implicated therapies that can directly remuscularize the heart without the need for cell transplantation, provided that the efficiency of reprogramming is sufficiently robust.
Here, we evaluated the efficiency of this direct cardiac reprogramming strategy using the GMT expression viruses reported in Ieda et al.10 and validated myocardial lineage reporters (αMHC-Cre, Nkx2.5-Cre, cTnT-Cre). We found a lack of αMHC or Nkx2.5 reporter activation in CFs and TTFs despite significant overexpression of GMT factors. However, with cTnT reporter we observed a ~35% labeling of fibroblasts that is confirmed by a ~250-fold increase in cTnT expression. However, the expression of other cardiac genes was minimally elevated. With GMT infection, we found 22% of infected fibroblasts exhibited a voltage-dependent calcium current without a spontaneous action potential, suggesting incomplete electrophysiological reprogramming. Furthermore, GMT-infected fibroblasts exhibited poor survival and minimal cardiac gene expression following transplantation into an injured murine heart in vivo. Together, our data suggest that direct cardiomyocyte reprogramming by GMT factors is inefficient and a greater understanding of transcription factor-mediated epigenetic change will be need to translate this promising approach into therapy.
The lentiviral tetracycline-inducible GMT expression vectors reported in Ieda et al.10 were kindly provided by Dr. Deepak Srivastava.
Detailed methods can be found in the Supplemental Materials.
To assess the efficiency of cardiomyocyte reprogramming from TTFs, we used αMHC-Cre/ROSA26mTmG mice that express membrane-tethered tandem dimerized Tomato (dTomato) at baseline, and switch to membrane-tethered enhanced GFP (eGFP) upon Cre-mediated excision in the ROSA locus (ROSA26mTmG) (Figure 1a,b)11. αMHC-Cre/ROSA26mTmG hearts are eGFP+ (Figure 1c) but TTFs are dTomato+ prior to GMT overexpression (Figure 1d). We infected freshly-isolated TTFs from αMHC-Cre/ROSA26mTmG mice with lentiviruses constitutively expressing rtTA along with doxycycline-inducible lentiviruses expressing Gata4, Mef2c, and Tbx5. Following induction with doxycycline for three weeks, we found no eGFP+ cells by immunofluorescence microscopy and flow cytometry with or without GMT lentiviral infection (Figure 1d).
To investigate this unexpected finding, we evaluated the induction of GMT overexpression in infected TTFs and found up to 1000-fold increases in GMT factors (Online Figure Ia). Although upregulation of Mef2c in infected TTFs remained modest (~10-fold) despite increases in viral titer, further investigation revealed that baseline levels of Mef2c in uninfected TTFs are already significantly elevated (Online Figure Id). Immunocytochemical staining for GMT proteins demonstrated their nuclear localization (Online Figure Ib). Luciferase reporter assays using enhancer/promoter elements previously described to report the transcriptional activities of Gata4, Mef2c, and Tbx5 proteins12–14 confirmed that each transcription factor is active in vitro (Online Figure Ic). To ensure that the ROSA26mTmG reporter can be efficiently excised by αMHC-Cre in vitro, we generated and differentiated αMHC-Cre/ROSA26mTmG ES cells and found robust expression of eGFP in beating cardiomyocytes (Online Figure II).
As αMHC is a marker of mature cardiomyocytes, we hypothesized that overexpression of developmentally essential genes Gata4/Mef2c/Tbx5 might induce an immature cardiac phenotype. We overexpressed GMT factors in TTFs from Nkx2.5 knock-in Cre/ ROSA26mTmG reporter mice, which express eGFP in immature cardiomyocytes (Figure 1a). No eGFP+ cells were detected after three weeks among infected fibroblasts (Figure 1e), suggesting a lack of Nkx2.5 upregulation.
We hypothesized that reprogramming CFs might be more efficient than reprogramming TTFs as CFs share developmental lineage history with cardiomyocytes. We overexpressed GMT in CFs from 2–3 week old αMHC-Cre/ROSA26mTmG mice and FACS-purified Thy1.2+/eGFP- CFs from Nkx2.5-Cre /ROSA26mTmG hearts. We first confirmed the upregulation of GMT in infected CFs (Online Figure Ie). Although Gata4 expression was increased by only 8-fold compared with uninfected CFs, we found a high baseline level of Gata4 expression in CFs (Online Figure If). Three weeks after GMT overexpression, we found no eGFP+ cells among either αMHC-Cre/ROSA26mTmG or Nkx2.5-Cre/ROSA26mTmG CFs (Figure 1f, g).
As neither αMHC- nor Nkx2.5 lineage-reporting fibroblasts conveyed cardiac reprogramming, we suspected that not all cardiac genes are equally induced by GMT overexpression. Quantitative PCR analysis of GMT-infected TTFs across a panel of cardiac genes confirmed the induction of some but not all cardiac genes (Figure 1h). Interestingly, while cTnT levels post-infection appeared modest compared with the high levels found in E10.5 hearts (Figure 1h), this represented a 250-fold increase in cTnT expression compared with uninfected TTFs (Online Figure III). In GMT-infected CFs, transcript levels of SERCA2a, Tbx20, and Gata6 were comparable to those in E10.5 cardiomyocytes (Figure 1i). Levels of cTnT, MyBPC and Gja1 also significantly increased but a number of important sarcomeric proteins failed to be induced.
Since cTnT was robustly upregulated by GMT factors, we overexpressed GMT in freshly-isolated TTFs from cTnT-Cre/ROSA26mTmG mice. Remarkably, up to ~35% of the cells became eGFP+ three weeks post-infection (Figure 2a). However, we noted that eGFP+ cells remained morphologically indistinguishable from eGFP- cells and exhibited no spontaneous beating activity (Figure 2b).
To further examine GMT-induced changes to gene expression on a genome-wide scale, we performed microarrays of CFs and TTFs before and after GMT infection. We selected for Tbx5 expressing cells by using a Tbx5-IRES-Puro lentivirus and treating the GMT infected cells with puromycin. Interestingly, we found no significant change in global gene expression profiles of CFs and TTFs after GMT overexpression (Online Figure IV). We noted, however, a subset of cardiac genes shifted towards cardiomyocyte-like expression patterns but these genes were either experimentally introduced (e.g. Tbx5) or known from the qPCR data above (e.g. cTnT) (data not shown).
While the global gene expression data shows a low overall efficiency of cardiomyocyte reprogramming by GMT factors, possibility remains that rare cells are more fully reprogrammed. To investigate this, we performed electrophysiological assessment of GMT infected TTFs at a single cell level. We compared GMT-overexpressing TTFs (n=32) with uninfected fibroblasts (n=26) and ES cell-derived cardiomyocytes (n=20) at three weeks post-infection. We found no spontaneous action potentials in GMT-infected (0/32 cells) or uninfected (0/26 cells) TTFs while ES cell-derived cardiomyocytes were all spontaneously active (20/20 cells) (data not shown). Upon pacing, ES cell-derived cardiomyocytes displayed typical murine cardiac action potentials, while uninfected fibroblasts demonstrated passive exponential decay of membrane potential consistent with a lack of active repolarization (26/26 cells) (Figure 2c). Interestingly, 7/32 (21.8%) of GMT-infected cells demonstrated up-sloping pacing induced action potential followed by passive exponential decay (Figure 2c).
We further examined this GMT-induced depolarization response in TTFs by introducing increasing stimulus amplitudes and found a graded response distinct from the “all or none” sodium current-dependent excitation typical of cardiomyocytes15 (Figure 2d). This absence of inward voltage-activated sodium currents in GMT-infected TTFs and their lack of active repolarization is likely responsible for their inability to fire repetitively upon high frequency pacing stimulation (data not shown). The ability of nifedipine, a dihydropyridine calcium channel antagonist, to block pacing induced action potentials (Figure 2e, red curve) revealed that the predominant component of these transient depolarizations was mediated by calcium and not sodium channels.
The low efficiency observed in GMT-overexpressing fibroblasts in vitro could have been explained by the absence of a supportive reprogramming environment. To examine the influence of a myocardial environment on reprogramming, we overexpressed GMT in CFs derived from transgenic mice that constitutively express luciferase and eGFP16 and injected these cells into the hearts of female SCID mice (5×105 cells/heart, n=3) that had just undergone surgical ligation of their left anterior descending (LAD) coronary arteries (Figure 3a). In parallel, uninfected cardiac-derived cells16 (5×105 cells/heart) were injected into the injured hearts of other SCID mice (n=3) as controls. Bioluminescence imaging over 8 days revealed a rapid loss of luciferase activity in hearts transplanted with GMT-infected CFs while only a modest degree of attrition was observed among uninfected cells (Figure 3b, c). To assess whether engrafted GMT overexpressing fibroblasts underwent cardiomyocyte reprogramming, we recovered transplanted single eGFP+ cells by FACS and evaluated their expression of a panel of cardiac genes using a novel Fluidigm® single cell PCR array. We found that recovered cells predominantly expressed vimentin, a marker of fibroblasts, while rare cells expressed a small number of cardiac genes (Figure 3d).
Direct cardiomyocyte reprogramming by overexpression of cardiac transcription factors is a conceptually appealing strategy for cardiomyocyte regeneration. Using transgenic mice expressing Cre recombinase under the regulation of αMHC, Nkx2.5 and cTnT promoters, we found that GMT overexpression in TTFs and CFs only induced expression of a subset of cardiac genes with minimal alteration of the fibroblast phenotype. We detected calcium channel-mediated depolarization currents in a subset of infected cells, suggesting that GMT reprogramming factors induced incomplete electrophysiological reprogramming. Transplantation of GMT-infected CFs into injured hearts resulted in no further improvements in the efficiency of cardiomyocyte phenotype conversion. Altogether, these data support a need for improved efficiency in cardiomyocyte reprogramming. A greater understanding of epigenetic changes associated with transcription factor overexpression will enhance the therapeutic potential of this approach.
Recent reports of direct reprogramming of fibroblast into other tissues such blood progenitors and neurons by overexpression of lineage-specific transcription factors17,18 offer hope that we may apply similar strategies to cardiac regenerative therapies. In our hands, however, the overall efficiency of cardiomyocyte reprograming with GMT overexpression is extremely low. Potential differences in experimental protocols (e.g. the method of fibroblast isolation, the method of virus production) or reagents used (e.g. genetic background of mouse strain, the cardiomyocyte-lineage reporters used) can influence the level of GMT overexpression and may account for some of the differences between our findings and those of Ieda et al.10 As an example, we found significant differences in the interpretations of reprogramming efficiency when different reporters (e.g. cTnT vs αMHC or Nkx2.5) are used. It is worth mentioning that the percentage of cTnT expressing cells in Ieda et al was only ~5% of the total infected cell population and among these, only a fraction of them are likely to express a more complete cardiac gene expression.
Our results highlight many challenges in transcription factor-based cardiac reprogramming. Importantly, we demonstrated the profound influences that choices of lineage reporters, cell types, and methods of evaluating cardiac phenotypes have on assessments of reprogramming efficiency. Moreover, our study raises important caveats for using GMT-reprogrammed fibroblasts as transplantable cardiomyocyte-like cells because these cells demonstrate poor survival post-transplantation (a common finding in previous cardiac transplantation experiments6) and are therefore unlikely to integrate with surrounding cardiomyocytes. Whether by adding different transcription factors and epigenetic modifiers to the GMT mix or by changing the starting cell type, significant improvements in the efficiency of cardiomyocyte reprogramming are needed before this strategy can be applied therapeutically.
The replacement of lost cardiomyocytes using more abundant cell types is an important goal in cardiac regenerative medicine. This study examines the efficiency of direct fibroblast reprograming into cardiomyocytes by the overexpression of GMT. In contrast to previous reports, it is found that GMT overexpression is inefficient at inducing a mature cardiomyocyte phenotype in TTF and CF. Furthermore, transplantation of GMT overexpressing CFs into injured murine hearts resulted in poor cell survival and minimal expression of cardiac genes. This study demonstrates the ongoing challenges in our ability to efficiently reprogram fibroblasts into cardiomyocytes for therapeutic applications.
We thank Dr. William Pu at Boston Children’s Hospital for cTnt-Cre/ROSAmtmg reporter fibroblasts; Dr. Robert Schwartz at University of Houston for Nkx2.5-Cre knock-in mice; Drs. X.J. Yang and Qi Kenneth Wang at Cleveland Clinic and Case Western University for promoter constructs used in luciferase reporter assays; Drs. Konrad Hochedlinger and Matthias Stadtfeld for the rtTA plasmid vector. Microarray studies were performed by the Molecular Genetics Core Facility at Children’s Hospital Boston supported by NIH-P50-NS40828, and NIH-P30-HD18655. Ms. Laura Prickett-Rice at the MGH-Center for Regenerative Medicine flow cytometry core facility provided assistance with FACS.
M. K. was supported by the German Research Foundation (KR3770/1-1). M.-A.D. was supported by the German Heart Foundation (Deutsche Herzstiftung e.V.). D.J.M was supported by NIH grants DA026982 and HL109004. S.M.W. was supported by NHLBI (HL081086, HL100408), NIH/OD (004411) and the Harvard Stem Cell Institute.